Formulating the machine learning problem, how research questions should be asked, and the trade-off of complexity versus accuracy (ARCHIVES S1E7)
Flirting with Models
00:00
Machine Learning
Using a deep neural network, there doesn't need to be any preemptive feature engineering. One of the added benefits of using this deep neural network is the opportunity to identify nonlinear features that very often we pre-engineer are linear in nature. The fitting of data is in fact a spectrum going from underfit to overfit. But somewhere in the middle there is like this Goldilocks point where you’re not too underfit and you're not too overfit.
Transcript
Play full episode
